Science Service System
You are here : Home : Proposals_Summary

Summary of Proposal LAN0464

TitleFire severity assessment and vegetation recovery monitoring in Mediterranean environments using TerraSAR-X, ALOS PALSAR and ERS data
Investigator Tanase, Mihai Andrei - University of Melbourne, Geomatics
Team Member
Prof. de la Riva, Juan - Universidad de Zaragoza, Geografia y Ordenacion del Territorio
Prof. Perez Cabello, Fernando - Universidad de Zaragoza, Geografia y Ordenacion del Territorio
Dr. Santoro, Maurizio - Gamma, Research & Development
Dr. Le Toan, Thuy - Centre d’Etudes Spatiales de la Biosphere, Physics and remote sensing of continental surfaces
SummaryIntroduction Throughout the last decades the forest management was influenced by the increase of forest fires and the variability in the response to fire over short-and longer time scales. With more than 10.000 fires per year Spain is one of Europe`s most affected countries. In recent years the effects of forest fires on environment dynamic in Aragon was studied with the help of remote sensing techniques based almost exclusively on optical sensors. The current project, intends to use synthetic aperture radar for detection of changes induced by fires and for post-fire monitoring of vegetation recovery. Objectives The research project aims to assess SAR effectiveness for evaluating changes induced by fires and the Mediterranean forest`s recover capacity through the integrated use of field and remote sensing data. The specific objective of this proposal is multi temporal analysis and characterization of fire affected areas using SAR data. The proposal is part of a project to evaluate the suitability of available spaceborne SAR sensors for fire severity assessment and vegetation recovery monitoring. SAR data of ERS1, ERS2, ASAR and JERS sensors is provided through the ESA project C1P.5446. PALSAR data is provided both through the above mentioned ESA project and ALOS Kyoto & Carbon Initiative. Finally, the current proposal would provide X-band SAR data of higher frequency and spatial resolution. Thus, data obtained by currently available SAR sensors bands (X-, C-, and L-) will be analyzed simultaneously for the same study area and conclusions on SAR data suitability for forest fire related studies could be draw for a typical Mediterranean environment. Methods Backscatter intensity and multi pass interferometric coherence and their temporal variations will be assessed for the retrieval of fire severity levels, biomass estimation and vegetation recovery monitoring in an area affected by fire in august 2008. Analysis: a) Compilation of available data (soil types, agricultural fields cartography, crop and vegetation types, forest inventory data, climatic data, digital elevation model, very high resolution optical imagery etc.) and field measurement of missing parameters (fire severity levels, biomass estimates for the selected parcels, etc.) b) Analysis of radar backscatter: X- band post fire backscatter response will be analyzed using SM and/or HS datasets (HH and HV polarizations) with respect to the fire severity levels assessed in the field. c) Interferometric analysis: multi pass coherence will be analyzed for fire severity levels evaluation and vegetation recovery after fire. Coherence`s evolution with respect to an image baseline will be also investigated. Pairs of Terrasar X tandem mission could be considered when/if available d) Evaluation of the relationships between radar measurements and indicators of fire severity and vegetation recovery speed provided by in situ estimations. Deliverables -Backscatter response of fire affected areas: seasonal /annual variations -Coherence properties of fire affected areas, evolution of coherence decorrelation subsequent to fires -Evaluation of the possibility to quantify fire effects and vegetation recovery using SAR data Funding Principal investigator is currently undertaking a PhD and has research interests in remote sensing of ecological parameters with a special interest in Mediterranean ecosystems. His studentship funded by the Spanish Ministry of Education and Science is supervised by Dr. Juan de la Riva and Dr. Thuy Le Toan. The field work is funded by the Spanish national research program through RS-Fire project (CGL2005-04863).
Final ReportThe most important findings of the study are resumed below: 1.Synthetic aperture radar systems have potential for burn severity estimation using either backscatter coefficient or the interferometric coherence. The potential of SAR data decreased with increasing frequency for both SAR metrics. The best results were obtained using cross-polarized channel in the case of backscatter and co-polarized channel in the case of coherence. 2.The joint use of the backscatter and the coherence did not improve significantly the estimation of burn severity in any of the radar frequencies tested. 3.The satellite viewing geometry (look angle) influenced the backscatter coefficient in form of an offset. However, the system's sensitivity to burn severity did not change significantly. 4.The correlation between the backscatter coefficient and burn severity increased with increasing spatial resolution, i.e. stronger speckle reduction after applying high multi-look factors. 5. The radar signal was strongly influenced by the topography. It is therefore imperative to take into account the local slope of the surface. Although in the case of interferometric coherence topography influence was lower, better results were obtained when the local incidence angle was considered. 6. A major error source for the estimation of burn severity was the environmental condition at image acquisition. In some specific cases and for certain polarizations precipitation could increase the sensitivity of SAR data to burn severity. However, the increased sensitivity was not always related to rainfall. Therefore, it is difficult to predict whether images acquired under wet condition would always represent the better choice especially after heavy precipitations. 7.Vegetation recovery monitoring was feasible using the backscatter coefficient of low-frequency bands (L). The high frequency bands (X and C) were less sensitive to changes in forest structure having a much lower potential for differentiating between regrowth stages. 8.Interferometric coherence (all frequencies) was almost insensitive to differences in forest structure except for recently burned areas where only low vegetation (e.g. grass and shrubs) was present. 9.Preliminary results from boreal forest confirmed the data trends obtained in the Mediterranean basin at least for specific environmental conditions (images acquired during the growing season in relatively dry conditions). The full report was uploaded as a publication

Back to list of proposals

© DLR 2004-2016